Source: Bruegel

Bruegel raises concerns that the EU’s forthcoming Code of Practice for AI may exacerbate existing legal tensions related to copyright protection. The policy draft, which aims to set voluntary industry standards for AI developers, has been criticized for failing to resolve ambiguities on training data and content use.

The think tank notes that the code lacks clear guidance on how AI companies should handle copyrighted material when training large models, potentially conflicting with EU copyright law and harming rights holders. Stakeholders worry that if adopted, the code may set poor precedents and provide a shield for non-compliance. Bruegel calls for stronger, binding legal provisions instead.